Ceevra Reveal 3+

Ceevra Reveal 3+ is software that processes CT and MR medical images to generate 3D visualizations for clinicians. It uses machine learning algorithms to create preliminary segmentations of normal anatomy (e.g., pulmonary arteries, kidneys, prostate) and enables interactive 3D image review, measurement, and manipulation for preoperative surgical planning and intraoperative display. The device is intended for use by healthcare professionals to assist clinical decision-making, with machine learning restricted to adult patients 22 years and older.

The updated device maintains the same fundamental features as the predicate: CT/MR image modalities, intended use by healthcare professionals in hospitals and clinics, Class II classification, interactive 3D manipulation and visualization, preoperative and intraoperative applications, and quantitative measurements (volume, diameter, distance). The key difference is that the updated device uses machine learning algorithms to generate semi-automated segmentations of pulmonary anatomy and other structures, whereas the predicate used non-machine-learning computer vision algorithms for this function.

IEC 62304:2006/Amd 1:2015 (Medical device software – Software life cycle processes); FDA Guidance documents on premarket submissions for software in medical devices and cybersecurity in medical devices. The machine learning models were verified using actual patient CT/MR imaging studies (133 total studies) with performance characterized by Sørensen–Dice coefficient (DSC) and Hausdorff distance metrics.

The indications for use are identical to the predicate. Although the updated device introduces machine learning technology for segmentation, it does not raise new safety or effectiveness questions because: (1) the overall intended use, user population, and clinical environment remain unchanged; (2) the same anatomical structures are segmented with high performance metrics (DSC ranging 0.82–0.93); (3) the machine learning models were validated against gold-standard medical professional segmentations using diverse patient populations and scanner manufacturers; (4) measurement accuracy was validated on both phantom and patient data; and (5) the device maintains the same output and clinical decision-support role as the predicate.